“Does Meta think we can’t read?” Meta abandons sketchy commercial terms update.

If you publish content linked from Facebook, you can continue to control AI access to it via robots.txt without needing to create test credentials.

Meta has indefinitely postponed a controversial update to its commercial terms that would have allowed the company to extensively train AI on content from external websites linked to Facebook pages. The proposed changes, which faced significant backlash from publishers, legal experts, and regulators, were initially set to take effect on October 30 but were later pushed to November 20. Critics warned that the vague language could override robots.txt directives, enabling Meta to bypass content owner preferences and potentially avoid licensing fees for AI training data. The update prompted legal action in Brazil and Norway, resulting in preliminary injunctions, and an investigation by the UK's Competition and Markets Authority. Meta cited a desire to engage with stakeholders and refine the policy as reasons for the delay, stating that existing terms remain in effect.

Meta's retreat signals that even tech giants face significant headwinds when attempting to unilaterally access vast datasets for AI training without explicit consent or compensation. The global regulatory and legal pushback demonstrates a growing consensus among authorities and industry bodies to protect intellectual property and user preferences in the age of generative AI. This event may embolden other content creators and publishers to challenge similar terms of service updates from AI companies.

The incident underscores the high stakes involved in data acquisition for AI development, highlighting the tension between platform ambitions and content owner rights. Future negotiations will likely see increased scrutiny on data scraping policies, potentially leading to more standardized opt-in mechanisms or licensing frameworks. Companies looking to build AI models will need to navigate this complex landscape with greater transparency and respect for existing content control protocols.